After Alphabet released its earnings, Google's stock plunged more than 10% in after-hours trading, nearly wiping out most of its gains over the past six months. Panic swept through the market, yet the underlying logic points to a valuation misfire during a transitional period rather than a collapse of the core business.
Slowing growth in traditional search advertising and negative free cash flow triggered a short-term selloff. However, Google Cloud's revenue doubling, the explosive growth of external TPU sales, and the top-ranking capabilities of the Gemini 4 model reveal strong support from new engines.
This massive divergence between short-term financial data and long-term strategic value is reminiscent of the similar market mispricing Microsoft experienced during the early stages of its cloud and AI transformation. Google is currently facing a similar "perception gap" test, and its long-term logic remains unchanged, further solidified by the accelerating rollout of new businesses. The market's lack of patience with the monetization cycle of B2B businesses led to this round of sharp price volatility, but this precisely provides long-term capital with a window to reassess the value of Google's full-stack AI closed loop.
Breaking down the focus of market panic, the pressure Google faced in the third quarter is indeed reflected in several key financial metrics falling short of expectations, but this is not a comprehensive stall of the core business. Alphabet's third-quarter total revenue reached $127.6 billion, up 25% year-over-year. While overall services continued to grow, it slightly missed the consensus estimate; advertising revenue was approximately $82 billion, also failing to meet Wall Street's high expectations.
The performance of the core advertising business is particularly critical. Search and other advertising revenue grew 12% year-over-year to $63.3 billion, not only below the market expectation of $64.9 billion but also far below the 14% growth floor set by buy-side institutions, compared to a 17% growth rate in the second quarter.
What worries the market more is management's guidance for the future. The cloud business, which just set a record high profit margin of 36.2%, is expected to face margin pressure in the fourth quarter due to increased use of third-party computing resources. On top of this year's capital expenditure of $195 billion to $205 billion, 2027 capital expenditure will "increase significantly." This means Google is simultaneously under triple pressure: slowing growth in the cash cow legacy business, delayed monetization of new AI models, and continued increases in infrastructure investment. The combination of these three constitutes a standard "valuation kill package" in the current US stock market, but the question is whether the market is overly fixated on current numbers while ignoring the structural reasons behind these pressures.
Digging deeper into the search business, the so-called "AI kills Google Search" remains a false premise; the truth lies in the misalignment between user behavior migration and monetization strategy. This quarter, Google's overall search query volume (including commercial queries) grew at the highest rate in the past five years. Under the AI mode, visual searches exceeded 50 billion per month, and the vast majority were "incremental searches" rather than simple replacements for traditional search. CEO Sundar Pichai repeatedly emphasized on the earnings call that, just like the shift to mobile a decade ago, Google focuses on building the best search user experience first and then monetizing it. Therefore, although the average AI query length has reached three times that of traditional search, monetization efficiency is relatively lower, and more interactions have not yet brought in advertising revenue comparable to traditional search.
Notably, positive signals have emerged in advertising experiments. Click-through rates for sponsored product listings and organic product listings in AI Mode are now basically equivalent. Google only recently announced the first international rollout of AI Mode ads and launched multi-turn conversation ads in the US, indicating its strategy is to first get users accustomed to the new search method and then gradually expand proven effective ad formats. As user habits migrate from "typing a few keywords" to "long-chain AI multi-turn interactions," the traffic pool has indeed grown larger, but Google has actively chosen to let monetization lag slightly to preserve entry-point stickiness. From a valuation perspective, AI search at this stage should not be seen as a burden dragging down performance but as an unfully realized "call option." If ad load, conversion rates, and revenue per query gradually improve, Google has every opportunity to unlock new revenue growth on top of its existing search ecosystem.
The sharp fluctuations in market sentiment largely stem from the conflict between Wall Street's lack of patience and CEO Sundar Pichai's engineer mindset. Current capital is extremely utilitarian, demanding immediate surges in quarterly EPS and advertising revenue. Especially against the backdrop of rising AI capital expenditure, every growth point not yet realized may be seen as evidence of declining future returns. Sundar Pichai himself has always been a typical engineer-minded CEO, emphasizing product experience, technological progress, infrastructure, and long-term returns, with an extreme lack of the gene for "painting big pictures" and inciting market sentiment with grand narratives. When pressed by analysts, he honestly explained the Q4 base pressure and computing power bottlenecks. Combined with the pressure of the legacy business slowing before new business monetization results are reflected in financial statements, the stock price naturally gave negative feedback on the spot. This conflict between short-term performance and long-term narrative makes it difficult for the market to understand Google's strategic resolve during the transition period, leading to a sharp pullback in the stock price after the earnings release. However, this pricing deviation based on short-term financial data often ignores the inevitable pains tech giants experience during transformation and the substantive breakthroughs in the underlying architecture of new businesses.
Shifting focus away from the somewhat fatigued legacy business, Google Cloud's surge has become the most dazzling part of Google's new engine. Its dual growth in revenue and profit, along with a massive backlog of orders, demonstrates the real strength of enterprise demand. In the third quarter, Google Cloud revenue surged 103% year-over-year to $30.7 billion, compared to approximately 82% growth in the previous quarter. Accelerating further on a hundred-billion-dollar scale itself indicates strong market demand.
More noteworthy is the profit performance. Google Cloud's operating income this quarter reached $11.1 billion, more than tripling year-over-year, with operating margin jumping from 23.7% in the same period last year to 36.2%, a record high. Meanwhile, the cloud business's unfulfilled backlog orders surged to $561 billion, with over 55% expected to be recognized as revenue within the next 24 months.
In terms of enterprise deployment, Google disclosed that sales of "knowledge work seats" across all its AI services have exceeded 100 million, and nearly 90% of Fortune 100 companies are using Gemini Enterprise. For example, BNP Paribas deployed the system for over 65,000 employees, and Tata Steel launched more than 300 specialized AI Agents within nine months to optimize supply chains and customer service. These cases prove that AI is truly taking over actual workflows within enterprises, moving beyond early demos and localized pilots. Once enterprises begin using AI for programming, knowledge work, data analysis, and business process automation, it will drive demand for computing power, storage, databases, and security services, helping Google turn AI into a complete set of infrastructure and services that can be continuously sold to enterprises.
Data compiled by Woofun AI shows that the truth behind the deteriorating free cash flow is not simply an "AI money-burning bottomless pit" but rather the rapid expansion of the external TPU sales business and the impact of one-time expenses. Many people see two consecutive quarters of negative free cash flow, far below expectations, and their first reaction is deteriorating fundamentals. But the breakdown by CFO Anat Ashkenazi and cloud business CEO Thomas Kurian reveals a different picture. In the third quarter, Alphabet reported free cash flow of negative $9.065 billion, further deteriorating from negative $5.855 billion in the second quarter, but this included a $5.2 billion final payment for the 2018 EU historic fine. Excluding this $5.2 billion EU fine cash payment, adjusted free cash flow was approximately negative $3.865 billion, actually an improvement of about $1.99 billion from the second quarter. Additionally, routine infrastructure capital expenditure was $51.4 billion, mainly used to support cloud business expansion and TPU inventory buildup.
The "negative $9.1 billion" free cash flow that scared the market actually obscures the enormous opportunity brought by external TPU sales. Google's TPU is transforming from an "internal chip used to save money" into a "platform product sold externally for computing power." This logic was fully realized into real orders in the third quarter, but the financial recognition cycle is completely different from traditional cloud leasing, resulting in massive upfront cash outflows. The explosion of external TPU sales marks the strategic success of Google's shift from internal chips to platform products. Inventory buildup is actually contract-driven stockpiling, not unsold goods. The latest training chip TPU 8T offers 2.7 times the performance of the previous generation, and the inference chip 8I improves cost-performance by 80%, providing strong pricing power in the current environment of scarce accelerated computing.
In terms of software ecosystem, Google partnered with Meta to launch TorchTPU, improving the training, inference, and deployment experience of PyTorch workloads on TPU, reducing developer migration costs. In terms of distribution channel expansion, TPU systems are no longer locked only in Google Cloud but are beginning to be deployed through partnerships with Oracle and Crusoe AI (a joint venture with Blackstone), and even directly into customers' on-premises data centers. Nearly 100 heavyweight customers (including Apple, Anthropic, Citadel, Jump Trading, etc.) are currently using TPU.
Why do TPU sales getting better make current cash flow look worse? Because the financial recognition cycle for selling complete TPU systems (including hardware, storage, optical interconnect networks, and software stacks) externally is different. Google disclosed on the earnings call: "All TPU system inventory corresponds to signed contracts; we only purchase components to build inventory when we have contracts." Customers have already signed binding contracts, and Google must first spend real money to buy parts and build inventory, which appears as a massive cash outflow on the third-quarter cash flow statement. It will only turn into massive revenue and cash flow when the systems are delivered at scale and billed in 2027. This is a cash flow cycle different from building its own data centers. Early inventory accumulation suppresses operating cash flow, and as systems are delivered and customers pay, the corresponding working capital occupation can gradually be released.
The top-ranking capabilities of the Gemini 4 series models fill in the most critical piece of Google's full-stack AI puzzle, and the surge in API calls validates the authenticity of market demand. Since the Gemini 4 series was unveiled, Google has not only held its ground but even surged to the front row in some key benchmarks. In addition to Gemini 4 Argon released two weeks ago, the latest Gemini 4 Carbon directly topped the global rankings with a score of 59 on the newly updated authoritative Artificial Analysis Intelligence Index, comprehensively outperforming Argon in full-stack web development, multimodal understanding, scientific reasoning, and cybersecurity. Combined with the newly released world model Genie 4, it further expands real-time interactive environment simulation capabilities, providing a foundation for robot and agent training.
Reflected in actual usage data, Google's first-party model API now processes nearly 32 billion tokens per minute, up 40% quarter-over-quarter and more than four times the same period last year. The Gemini app's user base has also officially crossed the 1 billion mark. The stronger the model capabilities and the lower the inference cost, the more opportunities Google has to deeply embed AI into Search, Workspace, and the Gemini app, and to further drive enterprise customers to use Google Cloud, call models, and deploy agents. Conversely, cloud infrastructure and self-developed TPU can support model training, inference, and large-scale deployment. This is the full-stack logic, where models, product entry points, cloud services, and underlying chips form a mutually reinforcing commercial closed loop. The model itself is not an isolated business but the core driving force of the entire ecosystem.
The core of the mispricing logic lies in the "perception gap" between B2B and B2C businesses, which causes the market to underestimate the value of long-term contracts. On one side is the search advertising business growing 12% but missing expectations, Google Cloud surging 103%, and a cloud business backlog of $561 billion. On the other side is a stock price that plunged as much as 10% after hours. This strong sense of disconnect inevitably brings to mind Microsoft's earlier experience. During Microsoft's critical transition to Azure cloud and early Copilot, Wall Street played out almost the exact same script: as long as the traditional Windows OEM business or personal computing segment slightly underperformed, or short-term capital expenditure surges suppressed free cash flow, the market would sell first regardless of Azure's surging underlying bookings. Google today faces a set of contradictions that are not identical but highly similar in underlying logic.
The rules of B2B business have always been "slow to warm up, lock in orders, recognize later." Whether enterprise customers are migrating workloads to Google Cloud, starting to use TPU in production environments, or willing to sign multi-year computing power contracts is difficult to judge from daily experience alone. The first step is a long internal validation process. For example, BNP Paribas deployed Gemini for 65,000 employees, and Tata Steel launched over 300 industrial Agents. During this period, the B2C market has almost no perception. The second step is order lock-in and upfront costs. Computing power and system demand are first locked into a backlog of up to hundreds of billions of dollars. To fulfill orders, Google even has to pay out of pocket to advance TPU inventory, enduring negative free cash flow pressure. The third step is the concentrated explosion of monetization. Only after computing power is delivered and Agents truly take over daily enterprise workflows, and customers move from "signing contracts" to "actually consuming tokens and continuously renewing," will revenue and profit be concentratedly released at some future point.
Reviewing Microsoft's history confirms that Google's long-term logic remains unchanged, and the current decline represents an opportunity for long-term capital. Sundar Pichai defined the upcoming 2027 as Google's "golden year of productivity," and the CFO also explicitly marked 2027 as the "harvest year" for large-scale TPU revenue recognition. Admittedly, slowing search advertising growth, negative free cash flow, and significantly increased 2027 capital expenditure are unavoidable realities and focal points the market will continue to debate in the fourth quarter. But if one reviews how Microsoft navigated the deep waters of "legacy business deceleration and new infrastructure buildup," perhaps one can gain a bit more conviction. When a giant with billions of C-end entry points, a large model back at the top, and $561 billion in B-end backlog orders is knocked into a 10% deep hole merely because of "actively delaying search monetization" and "cash flow timing differences from order-driven inventory buildup," this is very likely another mispriced chip handed to long-term capital by market sentiment under the "perception gap." Gemini 4.0's capability improvements, Google Cloud's 103% growth, and the explosion of external TPU sales together form a solid foundation for Google's AI counterattack. The current valuation adjustment is not a deterioration of fundamentals but a market misjudgment of the new business monetization cycle, providing long-term investors with an opportunity to reposition.
Comments